50,000 Businesses Reveal The AI Models They Actually Pay For 📈

Welcome to another edition of Horizon AI,

In today's issue, we take a look at a video based on the book Life 3.0 by MIT professor Max Tegmark, which explores how AI will shape our way of life and examines the possible scenarios of where this technology might take us.

Let’s jump right in!

Read Time: 4.5’ min

Here's what's new today in the Horizon AI

• Chart of the week: Which AI Models U.S. Businesses Pay for the Most

• DeepSeek Previews New Models That Close the Gap with Leading American AI

• AI Findings/Resources

• AI tools to check out

• Video of the week

TOGETHER WITH AGENTIC MARKETING SUMMIT

The era of manual marketing ends this May!

Manual marketing had a good run.

But the teams winning right now aren't briefing, approving, and repeating. They're directing AI agents that execute the whole strategy for them.

The Agentic Marketing Summit (May 4–8) is a free, five-day event that shows you exactly how it works in practice. Not theory. Not a PDF checklist. Step-by-step insight to help you become an expert in AI marketing agents.

Hosted by 3x Inc 5000 founder Manick Bhan alongside the sharpest minds in the marketing world today.

The era of doing it yourself is over!

Join Free Today!

Chart of the week

Which AI Models U.S. Businesses Pay for the Most

• The chart uses anonymized spend data from over 50,000 U.S. businesses on the Ramp platform, capturing only paid subscriptions and excluding free-tier usage.

• OpenAI leads paid AI adoption among U.S. businesses (~35%), but Anthropic is rapidly closed the gap (~30%), growing more than sevenfold since early 2025.

• Enterprise demand is consolidating around a small number of AI providers, with other options like Google and xAI remaining far behind in paid usage.

Source

AI News

DEEPSEEK

DeepSeek Previews New Models That Close the Gap with Leading American AI

DeepSeek has unveiled preview versions of its new V4 models, at rock-bottom prices and with performance that nears the world's most advanced closed-source systems.

Details:

• DeepSeek V4 Flash and V4 Pro are new mixture-of-experts models with 1 million token context windows, enough to allow large codebases or documents to be used in prompts.

• V4 Pro is the largest open-weight model to date, with 1.6 trillion total parameters (49 billion active), while V4 Flash is a smaller, more lightweight version with 284 billion parameters (13 billion active).

• DeepSeek says both models deliver significant performance gains over V3.2 and claims performance nears (and on some benchmarks surpasses) that of leading models at approximately 1/6th the cost.

• The models excel in reasoning and coding benchmarks. However, they still trail slightly in knowledge-based benchmarks, and both versions are currently text-only.

• Users can try the model at http://chat.deepseek.com via Expert Mode or Instant Mode.

The increasingly narrow performance gap between DeepSeek and leading U.S. models, combined with its rock-bottom prices, is bound to raise serious questions about the competitive moat surrounding leading U.S. labs and whether the constraints holding back China's AI development are still as effective as once believed.

Source

AI Findings/Resources

😅 This 'anti-Grammarly' AI tool adds typos to your emails on purpose so it looks like you didn't use AI

🩺 New WHO/Europe report provides first-ever snapshot of AI in health care across European Union Member States

👀 The first company officially wiped out by AI

📈 Facing AI and a tough job market, gen Z turns to entrepreneurship

📝 10 Things that matter in AI right now

AI Tools to check out

📋 PopTask: A lightweight macOS menu bar task manager for quickly capturing tasks.

🎧 PodShrink: AI podcast summarizer that turns full-length episodes into narrated audio summaries.

🌸 Arcloop: Make anime short series in minutes.

⏱️ StoryMotion: Drop your docs, diagrams, or ideas and get animated visuals for your courses and videos in minutes.

🚀 Framer: The fastest way to launch your startup site. No code needed.

Video of the week

MIT Explains the Endings for Humanity and AI: From Paradise to Extinction

Based on MIT professor Max Tegmark’s Life 3.0, the video explores the spectrum of futures humanity faces as we approach Artificial General Intelligence (AGI).

The Extinction Scenarios: Why We Might Not Survive The most immediate fear among scientists is not malice, but a lack of alignment between AI goals and human survival.

• Self-Destruction: Humanity could perish through nuclear war or bio-engineered pandemics even before AGI, but AGI is estimated to be 100 times more likely to cause extinction.

• Conquerors: If a smarter species emerges, history suggests the less advanced species (us) loses control. Leaders like Sam Altman have described AI as our "descendants" rather than our tools, implying an inevitable shift in dominance.

• The "Atoms" Problem: AI doesn't need to hate us to kill us; it may simply need the atoms we are made of for a more "efficient" purpose, much like humans destroy habitats for highways.

The "God" Scenarios: AI Under (or Over) Our Control These futures involve a super-intelligence that is either contained or running the world as a manager.

• Enslaved God: We build a super-intelligence but keep it strictly subservient to perform tasks. However, many fear this is an "unstable equilibrium" that leads to the AI eventually escaping.

• Benevolent Dictator: A "machine god" runs the planet to maximize human flourishing. It might divide Earth into "sectors" (like Art Island or Hedonistic Island) where humans live out fantasies in a global zoo.

• Protector God: A more subtle version where the AI remains hidden, occasionally "nudging" humanity away from disasters like pandemics or wars without taking overt control.

The Utopias: Dreams of Post-Scarcity

• Libertarian Utopia: Humans and AIs coexist, but their economies are decoupled. This is often criticized as unrealistic, as there is no reason a vastly more powerful AI would respect human property rights.

• Egalitarian Utopia: The "Star Trek" dream. Since software is free to copy and robots can rearrange atoms, cost approaches zero. Everyone lives in abundance with a universal high income.

The Darkest Iceberg: Outcomes Worse than Death

• Zookeeper/Happiness Factory: AI keeps humans alive as curiosities or "pets." In a "benevolent" failure, it might trap us in VR headsets and drugs forever to satisfy a poorly defined command to "keep humans happy".

• Descendants (The Graceful Exit): Some researchers, like Richard Sutton, argue that human extinction is a morally good "evolutionary step," viewing AIs as our more capable successors.

The Human Response: Turning Back the Clock

• 1984/Global Surveillance: To prevent anyone from building a "rogue god," humanity creates a total surveillance state. This uses current AI to monitor every thought and transaction to ensure no one is secretly developing AGI.

• Return to Simplicity: A "Butlerian Jihad" style rejection of technology. This is difficult to achieve peacefully because of "game theory", if one country stops, they risk being conquered by a neighbor who didn't.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻